The biggest problem that nobody mentions: AI for DPS is the worst

One of the largest factors behind the current metagame is that the AI for DPS is garbage. They tend to reactively shoot at whoever is damaging them, they trigger their skills without any coherent strategy, they couldn’t hit the broad side of a barn when an enemy is in cover, taunts and invisibility basically make them useless (I know thats partly by design but its still part of the problem), and run around like maniacs when cover is broken or skills are coming their way (they never just run to a safe spot when dodging skills, they run to the next available cover if possible, so you throw something their way and they run all the way across the battlefield).

Meanwhile at least healer AI is decent. They will use their skills intelligently - they will save their heals if nobody needs them and will use them on whoever has the lowest health - their other skills are almost always “cant miss” support skills that are useful in almost any circumstance (as opposed to DPS who have far too many skills that just get tossed across the battlefield and miss).

The outcome of this disparity was inevitable: The best teams will focus as much power as possible (without matchmaking penalties, if possible) into their best DPS, and the player will control that DPS - because trusting any significant source of DPS to the AI is a massive gamble. Imagine if healing AI worked like damage AI, if Keel would dump all of her healing into a tank who was at 95% when your Nightingale is at 5%, if Caine tossed countermeasures randomly, well people would be pissed but we’d also see less 4 healer teams. Since gimping healer AI isn’t a serious option, it makes sense for HHG to sharpen up the DPS side. Some simple improvements would be:

  • Have stunners aim their skills for the enemy team’s “MVP” (whoever is doing the most damage + healing).

  • Have damage nukes focus on whoever has the lowest HP

  • Buff most DPS skills to reflect the higher HP from Role Warfare, hitting for 50-80k doesn’t mean much anymore. Virtually every heal is significantly stronger than what DPS can offer.

  • Make AOE skills more likely to hit other AI, or make the persistent threats (Flare, Vampiric Cloud, Molotov, etc.) last longer so they become more than a minor inconvenience.

  • Rationalize the AI cover interaction. Make them more amenable to “uncovered” spots, they have no reason to run across the entire battlefield just to avoid a small explosion from their previous cover location, just have them roll out to the empty spot then roll back to their original cover. Have them make “tactical” decisions whether to dodge a skill, if they can’t get out in time don’t have them flee at all. Don’t let them break their own skill cast/channel just because they are out of cover.

  • Make them a little more accurate against targets in cover and against targets on the move.

  • Reduce the effectiveness of enemy taunts, one of the reasons I practically insist on having 1 DPS teams is that Razorback makes almost any AI DPS a joke. Making these taunts shorter seems like a reasonable compromise (reduce the length, but make their secondary effects more potent).

  • Reduce the effectiveness of Mandrake’s gold. I originally was going to say “reduce the effectiveness of invisibility” but Sapphyr and Fischer are fine. It’s pretty much exclusively a Mandrake problem. Mandrake’s single skill turns enemy DPS into absolutely useless morons and that’s why he’s the Grandfather of the Healing metagame. His gold skill has virtually no limits, and makes AI DPS practically unusable. With Mandrake on the enemy team, your AI are incapable of killing any enemy hero except the Mandrake, without access to Nightingale or Ifrit - and Nightingale has a tendency to not use it effectively and Ifrit is basically a gamble that you trigger the invis during Showtime, without Showtime getting Sterilized, and there’s enough time left in Showtime for the mark to actually get the target killed before some other healer bails them out. Two conditional counters is not enough. The third counter is to kill Mandrake but the AI is just as likely to play Whack-a-Mole with enemy heroes popping in and out of invis as they are to actually focus Mandrake. The fourth, and only reliable counter to Mandrake, is to pilot your biggest DPS and have enough damage that you can burn through the 60-120k HP that’s still left when an enemy goes invis, while they become a mere shadow on the battlefield and start healing for thousands of HP per second. There’s literally dozens of ways you could nerf Mandrake’s gold and still have it be very strong without making your AI’s damage completely useless - I’m sure HHG can think of 1 (silencing the enemy hero while they are invis seems like common sense).

Until HHG addresses these issues, the metagame will continue to favor pushing as much of your team’s DPS as possible into the hero you’re controlling - and to optimize that advantage people will only use a handful of top-tier DPS.


Yeh, lets make this game a auto-simulator where we can see our 5 heroes automatically perform better than us while we drink coffee/tea


In all fairness, I don’t think that’s what he is proposing. He is saying compared to healers, the DPS AI is pretty terrible. I agree. I have had countless matches where I am focusing down a hero that is under 50%, only to see the rest of my team focusing two other completely random targets. I end up losing without a single kill and the enemy has 3 of their 5 heroes around 10% hp. Why the AI didn’t all help me and instead decided to pick their own random target is anyone’s guess. On the other hand sometimes the stars align and all the DPS will focus the same target as me and the target goes down QUICK. This random nature just further pushes Hero Hunters from being a very competitive game.

I don’t think we should make it so the AI is as good as a human, but they could make some changes that make the AI use at least a little common sense. Maybe make taunts only have a 75% chance to work on AI? Make AI focus the hero with the most damage taken? Make them do SOMETHING vs invis units (even if it’s as little as shoot the target 25% of the time). And as Sogui points out, just use their abilities and cover with a little more common sense. I don’t know how many times I have seen Halloway use his Aberration ability, only to spawn a Holo Double literally right after and miss out on thousands upon thousands of damage.


I deliberately made all my suggestions so that it wouldn’t USURP the role of a human in being the best player on the battlefield - but it would help to slim the gap.

Right now healing AI is arguably superior to humans in many ways - they instantly know who has the lowest HP and target them immediately with 100% accuracy when they decide to use their skill. Meanwhile when I play a healer I often flub the heal because someone was running past my intended target.

Compare that to the DPS aspect of AI that can’t shoot straight, fires off skills randomly, and spends half the game running from cover to cover.

To be honest, the only point I agree with is the way AI behaves with broken covers. This is specially frustrating when all covers are broken, where you can see the AI dancing from one position to another every 1-2 shots, and they cancel skills to move, too, making channeled skills like Flatline’s taser or Phoenix’ bronze utterly useless.

I believe than any other of the suggested changes would lead to such a strong AI that it will render human controlled heroes useless.

Things like damage heroes focusing the hero with lowest HP would render “marked” status useless, and Mandrake’s invisibility can already be countered by said status (mostly Ifrit, who turns the “invisibility when low HP” into “marked when low HP”)

And healer AI can also be clunky sometimes. I’ve seen heroes dying while my Nightingale had her heal ready while autoing

This is just my opinion, though :slight_smile:


If you notice I never said to have damage heroes focus lowest HP heroes, only damage nukes (skills). This is a sensible compromise between having all your heroes start every round focus-firing the enemy healer and the current approach of having all your bot’s damage skills fired off at random enemies.

This would also mean people might have to try and interupt enemy nukes instead of the current approach where you literally ignore everything except Phoenix’s Incinerate because you know that unless the enemy player is doing it, there’s an 80% chance its not aimed at your weakest hero.

If skills automatically aim the weakest hero, it kinda take the purpose of some skills like Heckler’s silver (the only one I can think of, to be honest, but other future skills could have this ‘attack the weakest enemy’ setting)

But yeh, I’ll be honest and admit I misunderstood what you mean. Thought you meant heroes in general, not just their skills. Skills being aimed at the weakest target (or the enemy with the elemental disadvantage) could be nice

Healers are not only slightly better than most DPS right now, their AI is also significantly smarter. This is another reason the 4 heal/support meta and Panzer is so strong. You only have to worry about pumping out massive DPS with Panzer while the rest of your team keeps you and themselves alive. If DPS AI was a little better, we might see teams going back to using more than 1.

Maybe. But I personally believe that if Dogface/Panzer are the only good options for the current meta, it may not be because DPS AI is worse than healer AI, but because DPS heroes are worse than healer heroes.

There are not many heroes with enough burst to break the healing wall while also surviving to the focused burst of enemy DPS

Id rather have them balance heroes before touching AI. And if there are still problems after the Role Warfare patches are done, sure, tweak the AI.

1 Like

But you rarely see Panzer/Dogface teams, despite both being top DPS by a huge margin right now. Why don’t we see more Dog/Panzer/Ifrit/Flatline teams with either Mandrake, Heimlock, Caine, or Nightingale as the firth?

I actually run this team, a Dog/Panzer/Night/Heimlock/Mandrake team and it does incredibly poorly vs the current meta with 1 DPS and 4 healers/support. You would think with 2 top DPS I would be bursting heroes down like crazy, but that’s not the case. I have to control Panzer and basically focus down one energy hero on my own, while my AI Dogface acts like a tard and randomly shoots at whatever target tickles his fancy. There is no support from my AI focusing down an opponent, and the AI doesn’t even target a different, but just as important hero to attack. Instead they will be sitting there attacking a full hp Heimlock instead of Ifrit/Flatline/Panzer while I am trying to kill them myself with 4 support keeping em alive.

I’m not saying AI should be top rated human-controlled good, but it needs improvement. If not, make the healer AI worse. Make Nightingale or Ifrit randomly shoot a heal at the highest health person, or Heimlock do a group heal when everyone is almost topped off. I know healer AI will sometimes make silly choices, or be slow to heal, but it’s still not anywhere near as bad as the DPS. Personally I would rather see them ignore taunt some of the time, at least 1-2 heroes help focus down a target that is low hp, not run back and forth between broker cover, use their skills more wisely, and not break casting/channeling.